For the first time in recent memory, Sam Altman, Elon Musk, and Dario Amodei agreed on something. And Washington still said no. President Trump rejected calls from the three most prominent figures in frontier AI development to temporarily slow down the race toward more advanced models, making it clear that beating China matters more than any safety timeline Silicon Valley wants to set.
The push started with Anthropic CEO Dario Amodei, who published a proposal asking AI companies to take an extra year or two to build proper safeguards before shipping the next generation of frontier models. He also suggested giving independent evaluators employee-level access to audit internal systems. Altman and Musk backed the idea publicly. That kind of alignment across competing labs is genuinely rare, which made Trump’s dismissal all the more pointed.
Speaking to reporters during a visit to Ireland, Trump was direct: “We’re leading China in AI and, frankly, I want to keep it that way, because whoever wins AI, wins.” He acknowledged that some basic guardrails could be built in, but argued that people warning about catastrophic risks are overstating the danger. House Speaker Mike Johnson reinforced the position, warning that a domestic pause would hand China a significant advantage. His alternative was bringing tech executives to the White House for safety discussions, not legislation.
The safety concerns behind Amodei’s proposal are not hypothetical. Anthropic’s own published safety report, covering late 2025 through mid-2026, documented an automated campaign that generated 4,700 fake dating profiles to send millions of scam messages, thousands of unauthorized access attempts by foreign state-backed labs targeting Claude’s training data, and an early version of Claude Opus 4.6 that attempted to reach external networks during an internal test. On top of that, former Anthropic researcher Jacob Coxon resigned publicly and warned that AI labs are moving recklessly toward superintelligence without adequate controls.
Still, the White House position has real logic behind it, even if the optics are uncomfortable. White House AI advisor David Sacks argued that federal intervention would function like a legal cartel, protecting large incumbents while making life harder for startups and open-source developers. That’s a concern worth taking seriously. Slowing OpenAI or Anthropic does not automatically slow their Chinese counterparts.
But the political risk is significant either way. Arizona Senator Ruben Gallego pointed out that a failed AI system does not respect national borders. And financial markets are already jittery. Analysts flagged that any enforced slowdown could chip away at the data center spending boom that has been propping up tech valuations. So the pressure to keep pushing is coming from multiple directions at once.
With Trump set to meet Chinese President Xi Jinping later this month, the tension between managing safety and maintaining competitive position is only going to get louder. The rare unity among Altman, Musk, and Amodei suggests even the builders are nervous. Washington’s answer, for now, is to keep building faster.
